Conference Paper

Multiple Kernel Representation Learning for WiFi-Based Human Activity Recognition

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... A large number of researchers have begun to use information about the state of the communication channel to recognize human activity in the observed room [14][15][16][17][18][19][20][21]. The authors of [14] use a multi-core platform that automatically analyzes information about the state of the communication channel in a typical room and classifies human activity with an accuracy of up to 98%. ...
... A large number of researchers have begun to use information about the state of the communication channel to recognize human activity in the observed room [14][15][16][17][18][19][20][21]. The authors of [14] use a multi-core platform that automatically analyzes information about the state of the communication channel in a typical room and classifies human activity with an accuracy of up to 98%. The authors of [15] propose a system for detecting and recognizing human activity using information about the state of the communication channel. ...
... environments, and therefore it enables WiFi-based radar technology [2], [3]. WiFi-based radar can sense human motions by extracting CSI patterns by signal processing [4] or data-driven models [5], which has empowered many applications at smart homes including activity recognition [6], gesture recognition [7], human identification [8], [9], human-computer interface [10] and vital sign detection [4]. ...
... Update θ1, θ2, φ1, φ2 by minimizing Lp + λLm + γLg 6 where k denotes all categories. Note that the gesture or activity category, and even the recognition task can be customized by users. ...
Preprint
Full-text available
WiFi sensing technology has shown superiority in smart homes among various sensors for its cost-effective and privacy-preserving merits. It is empowered by Channel State Information (CSI) extracted from WiFi signals and advanced machine learning models to analyze motion patterns in CSI. Many learning-based models have been proposed for kinds of applications, but they severely suffer from environmental dependency. Though domain adaptation methods have been proposed to tackle this issue, it is not practical to collect high-quality, well-segmented and balanced CSI samples in a new environment for adaptation algorithms, but randomly captured CSI samples can be easily collected. In this paper, we firstly explore how to learn a robust model from these low-quality CSI samples, and propose AutoFi, an automatic WiFi sensing model based on a novel geometric self-supervised learning algorithm. The AutoFi fully utilizes unlabeled low-quality CSI samples that are captured randomly, and then transfers the knowledge to specific tasks defined by users, which is the first work to achieve cross-task transfer in WiFi sensing. The AutoFi is implemented on a pair of Atheros WiFi APs for evaluation. The AutoFi transfers knowledge from randomly collected CSI samples into human gait recognition and achieves state-of-the-art performance. Furthermore, we simulate cross-task transfer using public datasets to further demonstrate its capacity for cross-task learning. For the UT-HAR and Widar datasets, the AutoFi achieves satisfactory results on activity recognition and gesture recognition without any prior training. We believe that the AutoFi takes a huge step toward automatic WiFi sensing without any developer engagement while overcoming the cross-site issue.
... Channel State Information (CSI) shows how the WiFi signal encounters degradation due to the multipath when it travels from transmitter to the receiver [18]. These details are given at the granularity of the OFDM subcarriers. ...
Conference Paper
Full-text available
The building energy saving (BES) has been the subject of extensive research for reducing the energy consumption inside the buildings. One of the key solution for energy saving in buildings is to minimize the energy supply to the building areas that are not occupied by the inhabitants. However, this requires effective monitoring of occupants regardless of unpredictable variations in indoor environment, such as variation in the space size, furniture arrangement, nature of occupant’s activity (e.g., varied intensities and instances) etc. Currently, various occupancy monitory solutions have been employed in the existing smart buildings, namely PIR sensors, 𝐶𝑂2 sensors, cameras, etc. However, they are costly and sometimes not interoperable to the complex variations in indoor environments. In this paper, we leveraged the fine-grained information of physical layer (i.e., channel state information – CSI) of the commodity WiFi for occupancy detection and developed a self-adoptive method which is interoperable with complex variations in the indoor environment. In indoor contexts of different sized, varied intensities of physical activity, and various instances of activity of daily living (ADL), our testbed evaluation showed an average detection rate of 98.9%, 98.5%, and 98.1%, respectively.
... In contrast, by feeding significant amounts of data into machinelearning 25 or deep-learning models, 5,9 learning-based methods exhibit remarkable performance in complicated sensing tasks. Various deep neural networks have been designed to enable applications such as human activity recognition 26 and gesture recognition. 9 Although deep-learning models have performed admirably in function approximation, they require very high amounts of labeled data, which are expensive to collect and are adversely affected by distribution shifts induced by environmental dynamics. ...
Article
Full-text available
Over the recent years, WiFi sensing has been rapidly developed for privacy-preserving, ubiquitous human-sensing applications, enabled by signal processing and deep-learning methods. However, a comprehensive public benchmark for deep learning in WiFi sensing, similar to that available for visual recognition, does not yet exist. In this article, we review recent progress in topics ranging from WiFi hardware platforms to sensing algorithms and propose a new library with a comprehensive benchmark, SenseFi. On this basis, we evaluate various deep-learning models in terms of distinct sensing tasks, WiFi platforms, recognition accuracy, model size, computational complexity, and feature transferability. Extensive experiments are performed whose results provide valuable insights into model design, learning strategy, and training techniques for real-world applications. In summary, SenseFi is a comprehensive benchmark with an open-source library for deep learning in WiFi sensing research that offers researchers a convenient tool to validate learning-based WiFi-sensing methods on multiple datasets and platforms.
... The WiFi-based gait recognition method uses RF signals from WiFi-enabled devices to determine human identity. The transmitter emits WiFi signals, which are reflected by different body parts of the walking subject and then recorded by CSI data at the receiver [19], which has empowered many applications including occupancy detection [20], crowd counting [21], [22], human activity recognition [23], [24], [25], [26], [27], person identification [8], [28], vital sign detection [29], pose estimation [30] and gesture recognition [31], [32], [33]. To use WiFi sensing in the real world, current research aims at efficient communication [16], model security [34] and dataefficient training [35]. ...
Preprint
As an important biomarker for human identification, human gait can be collected at a distance by passive sensors without subject cooperation, which plays an essential role in crime prevention, security detection and other human identification applications. At present, most research works are based on cameras and computer vision techniques to perform gait recognition. However, vision-based methods are not reliable when confronting poor illuminations, leading to degrading performances. In this paper, we propose a novel multimodal gait recognition method, namely GaitFi, which leverages WiFi signals and videos for human identification. In GaitFi, Channel State Information (CSI) that reflects the multi-path propagation of WiFi is collected to capture human gaits, while videos are captured by cameras. To learn robust gait information, we propose a Lightweight Residual Convolution Network (LRCN) as the backbone network, and further propose the two-stream GaitFi by integrating WiFi and vision features for the gait retrieval task. The GaitFi is trained by the triplet loss and classification loss on different levels of features. Extensive experiments are conducted in the real world, which demonstrates that the GaitFi outperforms state-of-the-art gait recognition methods based on single WiFi or camera, achieving 94.2% for human identification tasks of 12 subjects.
... In contrast, by feeding a massive amount of data into machine learning [22] or deep learning networks, [9], [5], learning based achieve remarkable performances in complicated sensing tasks. Various deep neural networks are designed to enable many applications including activity recognition [23], gesture recognition [9], human identification [11], [12], [24], and people counting [13], [14]. Though deep learning models have a strong ability of function approximation, they require tremendous labeled data that is expensive to collect and suffer from the negative effect of distribution shift caused by environmental dynamics [25]. ...
Preprint
Full-text available
WiFi sensing has been evolving rapidly in recent years. Empowered by propagation models and deep learning methods, many challenging applications are realized such as WiFi-based human activity recognition and gesture recognition. However, in contrast to deep learning for visual recognition and natural language processing, no sufficiently comprehensive public benchmark exists. In this paper, we review the recent progress on deep learning enabled WiFi sensing, and then propose a benchmark, SenseFi, to study the effectiveness of various deep learning models for WiFi sensing. These advanced models are compared in terms of distinct sensing tasks, WiFi platforms, recognition accuracy, model size, computational complexity, feature transferability, and adaptability of unsupervised learning. It is also regarded as a tutorial for deep learning based WiFi sensing, starting from CSI hardware platform to sensing algorithms. The extensive experiments provide us with experiences in deep model design, learning strategy skills and training techniques for real-world applications. To the best of our knowledge, this is the first benchmark with an open-source library for deep learning in WiFi sensing research. The benchmark codes are available at https://github.com/xyanchen/WiFi-CSI-Sensing-Benchmark.
... This new tool improves the CSI resolution and enables CSI extraction from Atheros chip at WiFi Access Point (AP) instead of mini-PC for Intel 5300. Many WiFi sensing systems have been proposed for various applications including occupancy detection [4], crowd counting [16], human activity recognition [5], [17], [18], respiration monitoring [19], person identification [6] and gesture recognition [20], [3]. Recently, CSI can be extracted between a smartphone and a WiFi AP by Shadow Wi-Fi [21], which further increases the application range of WiFi sensing. ...
Preprint
Full-text available
WiFi technology has been applied to various places due to the increasing requirement of high-speed Internet access. Recently, besides network services, WiFi sensing is appealing in smart homes since it is device-free, cost-effective and privacy-preserving. Though numerous WiFi sensing methods have been developed, most of them only consider single smart home scenario. Without the connection of powerful cloud server and massive users, large-scale WiFi sensing is still difficult. In this paper, we firstly analyze and summarize these obstacles, and propose an efficient large-scale WiFi sensing framework, namely EfficientFi. The EfficientFi works with edge computing at WiFi APs and cloud computing at center servers. It consists of a novel deep neural network that can compress fine-grained WiFi Channel State Information (CSI) at edge, restore CSI at cloud, and perform sensing tasks simultaneously. A quantized auto-encoder and a joint classifier are designed to achieve these goals in an end-to-end fashion. To the best of our knowledge, the EfficientFi is the first IoT-cloud-enabled WiFi sensing framework that significantly reduces communication overhead while realizing sensing tasks accurately. We utilized human activity recognition and identification via WiFi sensing as two case studies, and conduct extensive experiments to evaluate the EfficientFi. The results show that it compresses CSI data from 1.368Mb/s to 0.768Kb/s with extremely low error of data reconstruction and achieves over 98% accuracy for human activity recognition.
... A general kernel method can be employed for recognition with local features. In several study [19]- [21] focuses on group activities where three different approaches are used to model person-person interaction. By exploring person-person interaction in the feature level for which a new feature representation called action contact (AC) descriptor is proposed. ...
Article
Full-text available
In recent times, fitness trackers and smartphones equipped with different sensors like gyroscopes, accelerometers, Global Positioning System sensors and programs are used for recognizing human activities. In this paper, the results collected from these devices are used to design a system that can have an application in monitoring a person’s health. Such systems take the raw sensor signals as input, preprocesses it and using machine learning techniques outputs the state of the user with minimum error. The objective of this paper is to compare the performance of different algorithms Logistic Regression, Support Vector Machine, k-Nearest Neighbor and Random Forest. The algorithms are trained and tested with an original number of features as well as with transformed number of features. The data with a smaller number of features is then used to visualize the high dimensional data. In this paper, each data point is mapped in the high dimensional data to two-dimensional data using t-distributed stochastic neighbour embedding technique. Overall, the first high dimensional data is visualized and compared with model’s performance with different algorithms and with different number of coordinates
... Many studies focus on how to process CSI data and most common solutions are based on machine learning algorithms, such as support vector machine(SVM) [14] and convolutional neural network(CNN) [15]. There are also some special solutions presented considering the characteristics of CSI, such as attention based bi-directional long short-term memory (ABLSTM) [16] and multiple Sensors 2021, 21, 2181 2 of 13 kernel representation learning (MKRL) framework [17]. In addition to studying which algorithm to use, studying how to apply WiFi based activity recognition to other fields is also a new and meaningful research direction. ...
Article
Full-text available
Currently, there are various works presented in the literature regarding the activity recognition based on WiFi. We observe that existing public data sets do not have enough data. In this work, we present a data augmentation method called window slicing. By slicing the original data, we get multiple samples for one raw datum. As a result, the size of the data set can be increased. On the basis of the experiments performed on a public data set and our collected data set, we observe that the proposed method assists in improving the results. It is notable that, on the public data set, the activity recognition accuracy improves from 88.13% to 97.12%. Similarly, the recognition accuracy is also improved for the data set collected in this work. Although the proposed method is simple, it effectively enhances the recognition accuracy. It is a general channel state information (CSI) data augmentation method. In addition, the proposed method demonstrates good interpretability.
... Channel State Information (CSI), extracted from the physical layer of wireless communications, leverages high resolution and makes it possible for human activity recognition [7] as well as more challenging works such as crowd counting [8], [9] and person identification [10], [11]. Current CSI-based activity recognition systems have three key limitations. ...
Preprint
In this study, we leveraged Channel State Information (CSI), commonly utilized in WLAN communication, as training data to develop and evaluate five distinct machine learning models for recognizing human postures: standing, sitting, and lying down. The models we employed were: (i) Linear Discriminant Analysis, (ii) Naive Bayes-Support Vector Machine, (iii) Kernel-Support Vector Machine, (iv) Random Forest, and (v) Deep Learning. We systematically analyzed how the accuracy of these models varied with different amounts of training data. Additionally, to assess their spatial generalization capabilities, we evaluated the models' performance in a setting distinct from the one used for data collection. The experimental findings indicated that while two models -- (ii) Naive Bayes-Support Vector Machine and (v) Deep Learning -- achieved 85% or more accuracy in the original setting, their accuracy dropped to approximately 30% when applied in a different environment. These results underscore that although CSI-based machine learning models can attain high accuracy within a consistent spatial structure, their performance diminishes considerably with changes in spatial conditions, highlighting a significant challenge in their generalization capabilities.
Article
Wi-Fi sensing technology has shown superiority in smart homes among various sensors for its cost-effective and privacy-preserving merits. It is empowered by channel state information (CSI) extracted from Wi-Fi signals and advanced machine learning models to analyze motion patterns in CSI. Many learning-based models have been proposed for kinds of applications, but they severely suffer from environmental dependency. Though domain adaptation methods have been proposed to tackle this issue, it is not practical to collect high-quality, well-segmented, and balanced CSI samples in a new environment for adaptation algorithms, but randomly captured CSI samples can be easily collected. In this article, we first explore how to learn a robust model from these low-quality CSI samples, and propose AutoFi, an annotation-efficient Wi-Fi sensing model based on a novel geometric self-supervised learning algorithm. The AutoFi fully utilizes unlabeled low-quality CSI samples that are captured randomly, and then transfers the knowledge to specific tasks defined by users, which is the first work to achieve cross-task transfer in Wi-Fi sensing. The AutoFi is implemented on a pair of Atheros Wi-Fi APs for evaluation. The AutoFi transfers knowledge from randomly collected CSI samples into human gait recognition and achieves state-of-the-art performance. Furthermore, we simulate cross-task transfer using public data sets to further demonstrate its capacity for cross-task learning. For the UT-HAR and Widar data sets, the AutoFi achieves satisfactory results on activity recognition and gesture recognition without any prior training. We believe that AutoFi takes a huge step toward automatic Wi-Fi sensing without any developer engagement. Our codes have been included in https://github.com/xyanchen/Wi-Fi-CSI-Sensing-Benchmark .
Article
As an important biomarker for human identification, human gait can be collected at a distance by passive sensors without subject cooperation, which plays an essential role in crime prevention, security detection and other human identification applications. At present, most research works are based on cameras and computer vision techniques to perform gait recognition. However, vision-based methods are not reliable when confronting poor illuminations, leading to degrading performances. In this paper, we propose a novel multimodal gait recognition method, namely GaitFi, which leverages WiFi signals and videos for human identification. In GaitFi, Channel State Information (CSI) that reflects the multi-path propagation of WiFi is collected to capture human gaits, while videos are captured by cameras. To learn robust gait information, we propose a Lightweight Residual Convolution Network (LRCN) as the backbone network, and further propose the two-stream GaitFi by integrating WiFi and vision features for the gait retrieval task. The GaitFi is trained by the triplet loss and classification loss on different levels of features. Extensive experiments are conducted in the real world, which demonstrates that the GaitFi outperforms state-of-the-art gait recognition methods based on single WiFi or camera, achieving 94.2% for human identification tasks of 12 subjects.
Article
As cameras and Wi-Fi access points are widely deployed in public places, new mobile applications and services can be developed by connecting live video analytics to the mobile Wi-Fi-enabled devices of the relevant users. To achieve this, a critical challenge is to identify the person who carries a device in the video with the mobile device's network ID, e.g., MAC address. To address this issue, we propose RFCam, a system for human identification with a fusion of Wi-Fi and camera data. RFCam uses a multi-antenna Wi-Fi radio to collect CSI of Wi-Fi packets sent by mobile devices, and a camera to monitor users in the area. With low sampling rate CSI data, RFCam derives heterogeneous embedding features on location, motion, and user activity for each device over time, and fuses them with visual user features generated from video analytics to find the best matches. To mitigate the impacts of multi-user environments on wireless sensing, we develop video-assisted learning models for different features and quantify their uncertainties, and incorporate them with video analytics to rank moments and features for robust and efficient fusion. RFCam is implemented and tested in indoor environments for over 800 minutes with 25 volunteers, and extensive evaluation results demonstrate that RFCam achieves real-time identification average accuracy of 97.01% in all experiments with up to ten users, significantly outperforming existing solutions.
Article
WiFi technology has been applied to various places due to the increasing requirement of high-speed Internet access. Recently, besides network services, WiFi sensing is appealing in smart homes since it is device-free, cost-effective and privacy-preserving. Though numerous WiFi sensing methods have been developed, most of them only consider single smart home scenario. Without the connection of powerful cloud server and massive users, large-scale WiFi sensing is still difficult. In this paper, we firstly analyze and summarize these obstacles, and propose an efficient large-scale WiFi sensing framework, namely EfficientFi. The EfficientFi works with edge computing at WiFi APs and cloud computing at center servers. It consists of a novel deep neural network that can compress fine-grained WiFi Channel State Information (CSI) at edge, restore CSI at cloud, and perform sensing tasks simultaneously. A quantized auto-encoder and a joint classifier are designed to achieve these goals in an end-to-end fashion. To the best of our knowledge, the EfficientFi is the first IoT-cloud-enabled WiFi sensing framework that significantly reduces communication overhead while realizing sensing tasks accurately. We utilized human activity recognition and identification via WiFi sensing as two case studies, and conduct extensive experiments to evaluate the EfficientFi. The results show that it compresses CSI data from 1.368Mb/s to 0.768Kb/s with extremely low error of data reconstruction and achieves over 98% accuracy for human activity recognition.
Article
Full-text available
Channel State Information (CSI) based human activity recognition has received great attention in recent years due to its advantages in privacy protection, insensitivity to illumination, and no requirement for wearable devices. In this paper, we propose a Multimodal Channel State Information Based Activity Recognition (MCBAR) system that leverages existing WiFi infrastructures and monitors human activities from CSI measurements. MCBAR aims to address the performances degradation of WiFi-based human recognition systems due to environmental dynamics. Specifically, we address the issue of non-uniformly distributed unlabelled data with rarely-performed activities by taking advantages of the generative adversarial network (GAN) and semi-supervised learning. We apply a multimodal generator to approximate the CSI data distribution in different environment settings with limited measured CSI data. The generated CSI data using the multimodal generator can provide better diversity for knowledge transfer. This multimodal generator improves the ability of MCBAR to recognize specific activities with various CSI patterns caused by environmental dynamics. Compared to state-of-the-art CSI-based recognition systems, MCBAR is more robust as it is able to handle the non-uniformly distributed CSI data collected from a new environment setting. In addition, diverse generated data from the multimodal generator improves the stability of the system. We have tested MCBAR under multiple experimental settings at different places. The experimental results demonstrate that our algorithm overcomes environmental dynamics and outperforms existing human activity recognition systems.
Article
Full-text available
Intelligent occupancy sensing is becoming a vital underpinning for various emerging applications in smart homes, such as security surveillance and human behavior analysis. However, prevailing approaches mainly rely on video camera, ambient sensors or wearable devices, which either requires arduous deployment or arouses privacy concerns. In this paper, we present a novel real-time, device-free and privacy-preserving WiFi-enabled IoT platform for occupancy sensing, which can promote a myriad of emerging applications. It is designed to achieve an optimal tradeoff between performance and scalability. Our system empowers commercial off-the-shelf (COTS) WiFi routers to collect Channel State Information (CSI) measurements and provides an efficient cloud server for computing via a lightweight communication protocol. To demonstrate the usefulness of our platform, an occupancy detection system is developed by exploiting the CSI curve of human presence. Furthermore, we also design an innovative activity recognition system based on our platform and machine learning techniques with high availability and extensibility. In the evaluation, the experimental results show that our platform enables these applications efficiently, with the accuracy of 96.8% and 90.6% in terms of occupancy detection and recognition respectively.
Article
Full-text available
Sedentary Behavior (SB) has been proved to be important risk factor for poor health, such as blood pressure and even cancer. However, existing sensor and vision-based SB detection approaches have limitations on practical usage and privacy concerns respectively. In this paper, we take the first attempt to develop a device-free SB monitoring and recommendation system namely CareFi, which leverages tremendous information behind WiFi signals to monitor the indoor environment and identify series of activities in SB. We deeply investigate the properties of channel state information and various activities in SB. According to different characteristics of static and dynamic activities, we design a foreground detection method to separate two categories and then adopt discriminative features of wireless signals in frequency and time domain to recognize them. Besides, we propose an update strategy to overcome the mutability of environment. We implement CareFi on Commercial The-Shelf (COTS) WiFi routers and evaluate its performance in both office and home environments. Experimental results demonstrate the robustness and accuracy of our method.
Article
Full-text available
Inferring abnormal glucose events such as hyperglycemia and hypoglycemia is crucial for the health of both diabetic patients and non-diabetic people. However, regular blood glucose monitoring can be invasive and inconvenient in everyday life. We present SugarMate, a first smartphone-based blood glucose inference system as a temporary alternative to continuous blood glucose monitors (CGM) when they are uncomfortable or inconvenient to wear. In addition to the records of food, drug and insulin intake, it leverages smartphone sensors to measure physical activities and sleep quality automatically. Provided with the imbalanced and often limited measurements, a challenge of SugarMate is the inference of blood glucose levels at a fine-grained time resolution. We propose Md3RNN, an efficient learning paradigm to make full use of the available blood glucose information. Specifically, the newly designed grouped input layers, together with the adoption of a deep RNN model, offer an opportunity to build blood glucose models for the general public based on limited personal measurements from single-user and grouped-users perspectives. Evaluations on 112 users demonstrate that Md3RNN yields an average accuracy of 82.14%, significantly outperforming previous learning methods those are either shallow, generically structured, or oblivious to grouped behaviors. Also, a user study with the 112 participants shows that SugarMate is acceptable for practical usage.
Article
Full-text available
Experiments show that operation efficiency and reliability of buildings can greatly benefit from rich and relevant datasets. More specifically, data can be analyzed to detect and diagnose system and component failures that undermine energy efficiency. Among the huge quantity of information, some features are more correlated with the failures than others. However there has been little research to date focusing on determining the types of data that can optimally support Fault Detection and Diagnosis (FDD). This paper presents a novel optimal feature selection method, named Information Greedy Feature Filter (IGFF), to select essential features that benefit building FDD. On one hand, the selection results can serve as reference for configuring sensors in the data collection stage, especially when the measurement resource is limited. On the other hand, with the most informative features selected by IGFF, the performance of building FDD could be improved and theoretically justified. A case study on Air Handling Unit (AHU) is conducted based on the dataset of the ASHRAE Research Project 1312. Numerical results show that, compared with several baselines, the FDD performances of conventional classification methods are greatly enhanced by IGFF.
Conference Paper
Full-text available
Pedestrian dead reckoning (PDR) is a promising complementary technique to balance the requirements on both accuracy and costs in outdoor and indoor positioning systems. In this paper, we propose a unified framework to comprehensively tackle the three sub problems involved in PDR, including step detection and counting, heading estimation and step length estimation, based on sequentially rotating the device (reference) frame to the Earth (reference) frame through sensor fusion. To be specific, a robust step detection and counting algorithm is devised according to vertical angular velocities and turns out to be tolerant of various smartphone placements; then, a zero velocity update (ZUPT) based algorithm is leveraged to calibrate the measurements in the Earth frame; on these grounds, the heading and step length are further estimated by exploiting the cyclic features of walking. A thorough and extensive experimental analysis is conducted and confirms the effectiveness and advantages of the proposed PDR framework as well as the corresponding algorithms.
Conference Paper
Full-text available
We propose Veto-Consensus Multiple Kernel Learning (VCMKL), a novel way of combining multiple kernels such that one class of samples is described by the logical intersection (consensus) of base kernelized decision rules, whereas the other classes by the union (veto) of their complements. The proposed configuration is a natural fit for domain description and learning with hidden subgroups. We first provide generalization risk bound in terms of the Rademacher complexity of the classifier, and then a large margin multi-ν learning objective with tunable training error bound is formulated. Seeing that the corresponding optimization is non-convex and existing methods severely suffer from local minima, we establish a new algorithm, namely Parametric Dual Descent Procedure (PDDP) that can approach global optimum with guarantees. The bases of PDDP are two theorems that reveal the global convexity and local explicitness of the parameterized dual optimum, for which a series of new techniques for parametric program have been developed. The proposed method is evaluated on extensive set of experiments, and the results show significant improvement over the state-of-the-art approaches.
Article
Full-text available
The location and contextual status (indoor or outdoor) is fundamental and critical information for upper-layer applications, such as activity recognition and location-based services (LBS) for individuals. In addition, optimizations of building management systems (BMS), such as the pre-cooling or heating process of the air-conditioning system according to the human traffic entering or exiting a building, can utilize the information, as well. The emerging mobile devices, which are equipped with various sensors, become a feasible and flexible platform to perform indoor-outdoor (IO) detection. However, power-hungry sensors, such as GPS and WiFi, should be used with caution due to the constrained battery storage on mobile device. We propose BlueDetect: an accurate, fast response and energy-efficient scheme for IO detection and seamless LBS running on the mobile device based on the emerging low-power iBeacon technology. By~leveraging the on-broad Bluetooth module and our proposed algorithms, BlueDetect provides a precise IO detection service that can turn on/off on-board power-hungry sensors smartly and automatically, optimize their performances and reduce the power consumption of mobile devices simultaneously. Moreover, seamless positioning and navigation services can be realized by it, especially in a semi-outdoor environment, which cannot be achieved by GPS or an indoor positioning system (IPS) easily. We~prototype BlueDetect on Android mobile devices and evaluate its performance comprehensively. The~experimental results have validated the superiority of BlueDetect in terms of IO detection accuracy, localization accuracy and energy consumption.
Conference Paper
Full-text available
WiFi based Indoor Positioning System (IPS) has become the most popular and practical system to provide Location Based Service (LBS) in indoor environments due to the availability of massive existing WiFi network infrastructures in buildings. WiFi based IPSs leverage received signal strengths (RSSs) from large numbers of WiFi access points (APs) and estimate the location of clients. Proper AP selection methods are required to select a subset of APs that provide useful information for localization in order to reduce the computational load and preserve or even enhance the localization accuracy of the entire IPS. Existing approaches select and measure the discriminative ability of APs individually during the offline phase without the consideration of the dependence among them. In this paper, based on mutual information of APs, we present online mutual information (OnlineMI), a novel AP selection strategy that measures the collective discriminative ability among APs. Furthermore, the proposed AP selection process is conducted online to adapt various environmental dynamics. Weighted k nearest neighbor (WKNN) approach is further employed as the localization algorithm after the OnlineMI AP selection strategy. Extensive experiments are carried out, and performance evaluation and comparison with existing methods demonstrates the superiority of the proposed OnlineMI-WKNN approach.
Article
Full-text available
Indoor Positioning System (IPS) has become one of the most attractive research fields due to the increasing demands on Location Based Services (LBSs) in indoor environments. Various IPSs have been developed under different circumstances, and most of them adopt the fingerprinting technique to mitigate pervasive indoor multipath effects. However, the performance of the fingerprinting technique severely suffers from device heterogeneity existing across commercial off-the-shelf mobile devices (e.g. smart phones, tablet computers, etc.) and indoor environmental changes (e.g. the number, distribution and activities of people, the placement of furniture, etc.). In this paper, we transform the Received Signal Strength (RSS) to a standardized location fingerprint based on the Procrustes analysis, and introduce a similarity metric, termed Signal Tendency Index (STI), for matching standardized fingerprints. An analysis on the capability of the proposed STI in handling device heterogeneity and environmental changes is presented. We further develop a robust and precise IPS by integrating the merits of both the STI and Weighted Extreme Learning Machine (WELM). Finally, extensive experiments are carried out and a performance comparison with existing solutions verifies the superiority of the proposed IPS in terms of robustness to device heterogeneity.
Conference Paper
Full-text available
We present the results, experiences and lessons learned from comparing a diverse set of technical approaches to indoor localization during the 2014 Microsoft Indoor Localization Competition. 22 different solutions to indoor localization from different teams around the world were put to test in the same unfamiliar space over the course of 2 days, allowing us to directly compare the accuracy and overhead of various technologies. In this paper, we provide a detailed analysis of the evaluation study's results, discuss the current state-of-the-art in indoor localization, and highlight the areas that, based on our experience from organizing this event, need to be improved to enable the adoption of indoor location services.
Article
Full-text available
Activity monitoring in home environments has become increasingly important and has the potential to support a broad array of applications including elder care, well-being management, and latchkey child safety. Traditional approaches involve wearable sensors and specialized hardware installations. This paper presents device-free location-oriented activity identification at home through the use of existing WiFi access points and WiFi devices (e.g., desktops, thermostats, refrigerators, smartTVs, laptops). Our low-cost system takes advantage of the ever more complex web of WiFi links between such devices and the increasingly fine-grained channel state information that can be extracted from such links. It examines channel features and can uniquely identify both in-place activities and walking movements across a home by comparing them against signal profiles. Signal profiles construction can be semi-supervised and the profiles can be adaptively updated to accommodate the movement of the mobile devices and day-to-day signal calibration. Our experimental evaluation in two apartments of different size demonstrates that our approach can achieve over 96% average true positive rate and less than 1% average false positive rate to distinguish a set of in-place and walking activities with only a single WiFi access point. Our prototype also shows that our system can work with wider signal band (802.11ac) with even higher accuracy.
Article
Full-text available
We introduce a framework of feature selection based on dependence maximization between the selected features and the labels of an estimation problem, using the Hilbert-Schmidt Independence Criterion. The key idea is that good features should be highly dependent on the labels. Our approach leads to a greedy procedure for feature selection. We show that a number of existing feature selectors are special cases of this framework. Experiments on both artificial and real-world data show that our feature selector works well in practice.
Conference Paper
Full-text available
Buildings are known to be the largest consumers of electricity in the United States, and often times the dominant energy consumer is the HVAC system. Despite this fact, in most buildings the HVAC system is run using primitive static control algorithms based on fixed work schedules causing wasted energy during periods of low occupancy. In this paper we present a novel control architecture that uses occupancy sensing to guide the operation of a building HVAC system. We show how we can enable aggressive duty-cycling of building HVAC systems - that is, turn them ON or OFF - to save energy while meeting building performance requirements using inexpensive sensing and control methods. We have deployed our occupancy sensor network across an entire floor of a university building and our data shows several periods of low occupancy with significant opportunities to save energy over normal HVAC schedules. Furthermore, by interfacing with the building Energy Management System (EMS) directly and using real-time occupancy data collected by our occupancy nodes, we measure electrical energy savings of 9.54% to 15.73% and thermal energy savings of 7.59% to 12.85% for the HVAC system by controlling just one floor of our four floor building.
Article
Full-text available
We are pleased to announce the release of a tool that records detailed measurements of the wireless channel along with received 802.11 packet traces. It runs on a commodity 802.11n NIC, and records Channel State Information (CSI) based on the 802.11 standard. Unlike Receive Signal Strength Indicator (RSSI) values, which merely capture the total power received at the listener, the CSI contains information about the channel between sender and receiver at the level of individual data subcarriers, for each pair of transmit and receive antennas. Our toolkit uses the Intel WiFi Link 5300 wireless NIC with 3 antennas. It works on up-to-date Linux operating systems: in our testbed we use Ubuntu 10.04 LTS with the 2.6.36 kernel. The measurement setup comprises our customized versions of Intel's close-source firmware and open-source iwlwifi wireless driver, userspace tools to enable these measurements, access point functionality for controlling both ends of the link, and Matlab (or Octave) scripts for data analysis. We are releasing the binary of the modified firmware, and the source code to all the other components.
Article
Accidental falls have been identified as a cause of mortality for elders who live alone around the globe. Following a fall, additional injury can be sustained if proper fall recovery techniques are not followed. These secondary complications can be reduced if the person had access to safe recovery procedures or were assisted, either by a person or a robot. We propose a framework for in situ robotic assistance for post fall recovery scenarios. In order to assist autonomously robots need to recognize an individual's posture and subactivities (e.g., falling, rolling, move to hands and knees, crawling, and push up through legs, sitting or standing). Human body skeleton tracking through RGB-D pose estimation methods fail to identify the body parts during key phases of fall recovery due to high occlusion rates in fallen, and recovering, postures. To address this issue, we investigated how low-level image features can be leveraged to recognize an individual's subactivities. Depth cuboid similarity features (DCSFs) approach was improved with M-partitioned histograms of depth cuboid prototypes, integration of activity progression direction, and outlier spatiotemporal interest point removal. Our modified DCSF algorithm was evaluated on a unique RGB-D multiview dataset, achieving 87.43 ± 1.74% accuracy in the extensive 3003 (C¹⁵/10 ) combinations of training-test groups of 15 subjects in 10 trials. This result was significantly larger than the nearest competitor, and faster in the training phase. This work could lead to more accurate in situ robotic assistance for fall recovery, saving lives for victims of falls.
Conference Paper
Power delay profiles characterize multipath channel features, which are widely used in motion- or localization-based applications. Recent studies show that the power delay profile may be derived from the CSI traces collected from commodity WiFi devices, but the performance is limited by two dominating factors. The resolution of the derived power delay profile is determined by the channel bandwidth, which is however limited on commodity WiFi. The collected CSI reflects the signal distortions due to both the channel attenuation and the hardware imperfection. A direct derivation of power delay profiles using raw CSI measures, as has been done in the literature, results in significant inaccuracy. In this paper, we present Splicer, a software-based system that derives high-resolution power delay profiles by splicing the CSI measurements from multiple WiFi frequency bands. We propose a set of key techniques to separate the mixed hardware errors from the collected CSI measurements. Splicer adapts its computations within stringent channel coherence time and thus can perform well in presence of mobility. Our experiments with commodity WiFi NICs show that Splicer substantially improves the accuracy in profiling multipath characteristics, reducing the errors of multipath distance estimation to be less than $2m$. Splicer can immediately benefit upper-layer applications. Our case study with recent single-AP localization achieves a median localization error of $0.95m$.
Conference Paper
Some pioneer WiFi signal based human activity recognition systems have been proposed. Their key limitation lies in the lack of a model that can quantitatively correlate CSI dynamics and human activities. In this paper, we propose CARM, a CSI based human Activity Recognition and Monitoring system. CARM has two theoretical underpinnings: a CSI-speed model, which quantifies the correlation between CSI value dynamics and human movement speeds, and a CSI-activity model, which quantifies the correlation between the movement speeds of different human body parts and a specific human activity. By these two models, we quantitatively build the correlation between CSI value dynamics and a specific human activity. CARM uses this correlation as the profiling mechanism and recognizes a given activity by matching it to the best-fit profile. We implemented CARM using commercial WiFi devices and evaluated it in several different environments. Our results show that CARM achieves an average accuracy of greater than 96%.
Article
Offering a fundamental basis in kernel-based learning theory, this book covers both statistical and algebraic principles. It provides over 30 major theorems for kernel-based supervised and unsupervised learning models. The first of the theorems establishes a condition, arguably necessary and sufficient, for the kernelization of learning models. In addition, several other theorems are devoted to proving mathematical equivalence between seemingly unrelated models. With over 25 closed-form and iterative algorithms, the book provides a step-by-step guide to algorithmic procedures and analysing which factors to consider in tackling a given problem, enabling readers to improve specifically designed learning algorithms, build models for new applications and develop efficient techniques suitable for green machine learning technologies. Numerous real-world examples and over 200 problems, several of which are Matlab-based simulation exercises, make this an essential resource for graduate students and professionals in computer science, electrical and biomedical engineering. Solutions to problems are provided online for instructors.
Fall recovery subactivity recognition with rgb-d camerasDutycycling buildings aggressively: The next frontier in hvac control
  • K I Withanage
  • I Lee
  • R Brinkworth
  • S Mackintosh
  • D Thewlis Agarwal
  • B Balaji
  • S Dutta
  • R K Gupta
  • T Weng
K. I. Withanage, I. Lee, R. Brinkworth, S. Mackintosh, and D. Thewlis, "Fall recovery subactivity recognition with rgb-d cameras," IEEE Transactions on Industrial Informatics, vol. 12, no. 6, pp. 2312-2320, 2016. [2] Y. Agarwal, B. Balaji, S. Dutta, R. K. Gupta, and T. Weng, "Dutycycling buildings aggressively: The next frontier in hvac control," in Information Processing in Sensor Networks (IPSN), 2011 10th International Conference on, pp. 246-257, IEEE, 2011.